Using RSEM expected counts in DESeq2

[thileepan sekaran]

Hi Brian and others, 

In the Trinity pipeline incorporates RSEM and other alternatives(kallisto, salmon,eXpress) for estimation, the expected counts is later rounded off before proceeding into further steps in DESeq2. 

I found that there is a library called tximport(in bioconductor since 6 months)which can summarize the output from RSEM, kallisto, salmon, sailfish and later can be imported into DESeq2 pipeline using DESeqDataSetFromTximport() function. Detailed explanation of using RSEM expected count sin DESeeq2 is provided by Michael Love(one of the author DESeq2) is given int he following link: https://bioconductor.org/packages/devel/bioc/vignettes/tximport/inst/doc/tximport.html.

I would like to know whether this approach of using tximport() and DESeqDataSetFromTximport() is better than rounding off? If yes, will this method be incorporated in further releases? Kindly guide me.

Regards

Dinesh

[Brian Haas]

Hi Dinesh,

Yes, using Tximport() is a preferred approach and we'll integrate this into a future version of Trinity.  I seriously doubt it'll make much of a difference since it mostly addresses certain edge-case scenarios in a theoretical sense, but it would be good to leverage the best approaches where possible and to keep current.

~b

--

--
Brian J. Haas
The Broad Institute
http://broadinstitute.org/~bhaas

 

[Pablo García Fernández]

Hi Brian,

I used tximport and then I went in to DESeq2 but I had to admit it I find DESeq2 easy to use but in order to obtain the output as organized as your Trinity scripts do... is not so easy... (at least for me, low level user of R...)

So my question is, how can I combine tximport output and Trinity scripts for DGE:
 

$TRINITY_HOME/Analysis/DifferentialExpression/run_DE_analysis.pl 
$TRINITY_HOME/Analysis/DifferentialExpression/analyze_diff_expr.pl

For the first one I think I can just substitute --matrix XYZ.counts.matrix for the matrix obtained when I ran tximport. However, in the second one the input matrix should be a TMM.EXPR.matrix. Do I just follow the same kind of normalization over my first tximport report?

I'm quite insecure now about the next steps and I don't want to introduce an unexpected fountain of error.

Thank you,

Pablo GF

[b.aza...@gmail.com]

سلام

شما ایران زندگی می کنید؟ من به کمک یک فارس زبان نیاز دارم اگگه جواب بدید خیلی ممنون میشم.